Uncertainty and Decision Making for Community Scale Urban Stormwater Modeling
Date:
Abstract: Stormwater modeling software are increasingly used to plan for and implement long-lasting infrastructure in urban areas. The EPA’s Storm Water Management Model (SWMM) is a well-documented and widely used modeling software. SWMM and other related software, such as PCSWMM, must be calibrated and validated before using them to inform any decisions, and this process must consider the multiple sources of uncertainty in model inputs, model parameters, and model structure. Many models in the literature are calibrated to a single or a few events, but recent research has shown that models calibrated for one set of events will not guarantee good performance of the model under another set of events. Our research demonstrates how different modeling decisions can affect performance of a 1D dual drainage stormwater model in PCSWMM representing storms observed over 2022 in a community in Pittsburgh, PA. Explicit and implicit modeling decisions such as the flow metric of interest, the observed storms used for calibration, the location(s) where the model is calibrated, and the spatial and temporal resolution of rainfall data can introduce hidden uncertainties when planning for stormwater infrastructure. These decisions can be further complicated by the deep uncertainty associated with climate change. We offer suggestions and recommendations for modelers to be aware of and appropriately quantify these uncertainties to support responsible and equitable stormwater modeling under deep uncertainty.